pandas.read_csv在标题下放置行时导致列标签移位

Question

I am trying to read a .csv file with pandas, with a header looking like this:

System Information_1
System Information_2
System Information_3
System Information_4

"Label1"; "Label2"; "Label3"; "Label4"; "Label5"; "Label6"
"alternative Label1"; "alternative Label2"; "alternative Label3"; "alternative Label4"; "alternative Label5"; "alternative Label6"
"unit1"; "unit2"; "unit3"; "unit4"; "unit5"; "unit6"

I'm using the following code to read it:
df = pd.read_csv('data.csv', sep=';', header=5, skiprows=[6,7], encoding='latin1')

My dataframe does however end up having "unit1", "unit2", "unit3", "unit4", "unit5", "unit6" instead of "Label1", "Label2", "Label3", "Label4", "Label5", "Label6" as column labels.

In an older version of my csv-file, however, the import code works properly. The difference I could spot between the files was that the older file has a full set of separators in the first 4 rows:

System Information_1;;;;;
System Information_2;;;;; 
etc.  

Does anyone know where that error comes from and how to solve it?

Answer 1

您也可以跳过第一行，但是也不要将header设置为5 ，因为那时它是0，因此可以将其保留为自动检测：

df = pd.read_csv('data.csv', sep=';', skiprows=[0,1,2,3,4,6,7], encoding='latin1')

Answer 2

You could use a list as your header argument:

import pandas as pd
from io import StringIO

data = """System Information_1
System Information_2
System Information_3
System Information_4

"Label1"; "Label2"; "Label3"; "Label4"; "Label5"; "Label6"
"alternative Label1"; "alternative Label2"; "alternative Label3" "alternative Label4"; "alternative Label5"; "alternative Label6"
"unit1"; "unit2"; "unit3"; "unit4"; "unit5"; "unit6" 
1;2;3;4;5;6
10;20;30;40;50;60
"""

df = pd.read_csv(StringIO(data), sep=';', header=[4], skiprows=[6, 7], encoding='latin1')

gives:

Answer 3

The "header" parameter starts counting after the "skiprows" parameter.

If you want to use the label as header:

df = pd.read_csv('pruebasof.csv', sep=';', skiprows=[0,1,2,3,4,6], encoding='latin1')

Otherwhise, if you want to use the alternative label as header:

df = pd.read_csv('pruebasof.csv', sep=';', skiprows=6, encoding='latin1')

I made it so you can use the label while keeping the "units" as data for the labels.